Dual Solenoid PID Idle

[mattias]

I will make an effort to update the "F1 for help" dialogs, they are messed up right now since it has been split into four separate dialogs (used to be two, maybe three).

[rob@vems.co.uk]

I think you could ask for help on this - if needs be we could make a section in here to place information/instructions, then it would be a case of cutting the words from here and pasting them into the help files...

[mattias]

I wrote a bunch earlier, so I'll start by just moving it to the correct dialog. Then I'll input new things from the wiki links in my earlier post (bottom first page in this thread).

[Benzmac16v]

You guys are awesome!

I got it working decent enough for hot temps (>70°C).  It is a little over damped but that is ok for now as I do not get those annoying drops that like to stall the car when I take my foot off the pedal.

It was idling higher than I thought it should be when I enabled PID and the IAC int value was very negative.  It was idling about 50-150rpm higher than without PID control. The engine temp may have increased.  I assume that VEMS will scale the iacv % in between the temp ranges.  But does it do this after your peak temp? These cars apparently run hot... VEMS says its about 100°C, stock gauge says 80°C, and I double check with infrared and a probe on the tstat housing which both said about 100°C.  I haven't tuned the cell before my highest temp and it is lower, so if VEMS continues to scale it would have been opening my iacv.

I will upgrade eventually, but I am trying to take it slow and learn one thing at a time.  I looked into it (and tried it) a while ago, but there wasn't an easy way to import my settings and maps from MS to VEMSTune (at least that I could find) so I just gave up and figured my problems were more my lack of understanding rather than an older firmware version (which was true).  But I also know that a lot of the code got much better and there are a few things worth upgrading for (idle, acceleration enrich) but ill get there.

Thanks,
Jim

[Benzmac16v]

I think I have it all working now, Thank you guys very much.  It is very stable now.  The IAC INT is still very large, around 100-120.  The odd thing about it (at least to me) is that it doesn't really move much, it only changes when I enter idle.  It does not continue to react with my idle speed.  When I had the PI set too high and the idle would oscillate, the IAC INT stayed constant.  Is that normal behavior?

Jim

[gunni]

Lower the integral decrease and increase limits, the close IACint is to 0 the faster it will react.

[mattias]

If the "IAC int" is far from 0, then the reference table is not setup correctly.

When the table is spot on, you have "IAC int" = 0, in an ideal world.

This is not so, it's better to have a negative "IAC int" so that the integral closes the valve slightly to maintain the idle rpm, making it stronger instead of dropping below the target rpm before the algorithm takes over.

Jim : The IAC integral is only supposed to do any work when you are below the "TPS idle threshold", it's supposed to only change/work when you're at idle.

[Benzmac16v]

I know it should only change when the car is idling.  However, it only changes once when it enters idle.  Even though the idle will fluctuate slightly.  Even if I increase the PI values to make the system unstable, it still will only update the IAC INT once.

I am fairly confident that I have a good reference table (at least when warm) because it idles very well, and idles slightly higher than my target if I disable PID control.

Also, my increase decrease limits are very low.  Increase is at 20 and decrease is at 15.

Jim

[mattias]

I think you should use a lot higher limits for increase/decrease, the integral is most likely limited.

[Benzmac16v]

Ok, so I have continued playing and found some odd behavior.

Basically, if I jack up the number in my ref table (to say, 70%) and put my inc/dec limits to 255, when I enter idle it will bounce a lot, but the IAC int number will decrease and close in on zero, the idle will also begin to settle.

So I have the video attached of the megatune output.  I also have the log files.

Video: http://www.youtube.com/watch?v=P2r8MZ68ZE0 *EDIT* Youtube butchered the quality, but the top left gauge is RPM, bottom right is IAC int. Ill see if I cannot make it better...  should be good now

Log of the 'Good' IAC int
http://www.rpi.edu/~smithj16/Car%20Vids/GoodIACint.xls

Here is the log of the, in my opinion, good idle
http://www.rpi.edu/~smithj16/Car%20Vids/goodIdle.xls

Tell me what you think,
Jim

[mattias]

 I have not looked at your logs.

Don't use ref dc values that don't make sense. We/you don't have any idea of the operating range of the valve. Clear the PID regulator so that it doesn't work and create a good ref dc table that puts it at 200 rpm above what you want at each coolant temp with all you normally have turned on. Then you can adjust the rest.

Ideally we would have a separate ignition advance for just idle, so that the main ignition table doesn't affect the idle. For now, just make sure you have the same advance at all idle load sites - that helps a lot with getting this adjusted.

[Benzmac16v]

Well it is a gradual thing.  I used 70 only because it makes it so clear.  If I just increase the ref value little by little the IAC int gets 'better'.  A value of about 36 works very good, but I get an IAC int of about 110-120.

Jim